Owing to the variable shapes, large size difference, uneven grayscale and dense distribution among biological cells in an image, it is still a challenging task for the standard Mask R-CNN to accurately detect and segment cells. Especially, the state-of-the-art anchor-based methods fail to generate the anchors of sufficient scales effectively according to the various sizes and shapes of cells, thereby hardly covering all scales of cells.
We propose an adaptive approach to learn the anchor shape priors from data samples, and the aspect ratios and the number of anchor boxes can be dynamically adjusted by using ISODATA clustering algorithm instead of human prior knowledge in this work. To solve the identification difficulties for small objects owing to the multiple down-samplings in a deep learning-based method, a densification strategy of candidate anchors is presented to enhance the effects of identifying tinny size cells. Finally, a series of comparative experiments are conducted on various datasets to select appropriate a network structure and verify the effectiveness of the proposed methods.
The results show that the ResNet-50-FPN combining the ISODATA method and densification strategy can obtain better performance than other methods in multiple metrics (including AP, Precision, Recall, Dice and PQ) for various biological cell datasets, such as U373, GoTW1, SIM+ and T24.
Our adaptive algorithm could effectively learn the anchor shape priors from the various sizes and shapes of cells. It is promising and encouraging for a real-world anchor-based detection and segmentation application of biomedical engineering in the future.
Our adaptive algorithm could effectively learn the anchor shape priors from the various sizes and shapes of cells. It is promising and encouraging for a real-world anchor-based detection and segmentation application of biomedical engineering in the future.Pneumonia is a disease that affects the lungs, making breathing difficult. Nowadays, pneumonia is the disease that kills the most children under the age of five in the world, and if no action is taken, pneumonia is estimated to kill 11 million children by the year 2030. Knowing that rapid and accurate diagnosis of pneumonia is a significant factor in reducing mortality, acceleration, or automation of the diagnostic process is highly desirable. The use of computational methods can decrease specialists' workload and even offer a second opinion, increasing the number of accurate diagnostics.
This work proposes a method for constructing a specific convolutional neural network architecture to detect pneumonia and classify viral and bacterial types using Bayesian optimization from pre-trained networks.
The results obtained are promising, in the order of 0.964 accuracy for pneumonia detection and 0.957 accuracy for pneumonia type classification.
This research demonstrated the efficiency of CNN architecture estimation for detecting and diagnosing pneumonia using Bayesian optimization. The proposed network proved to have promising results, despite not using common preprocessing techniques such as histogram equalization and lung segmentation. This fact shows that the proposed method provides efficient and high-performance neural networks since image preprocessing is unnecessary.
This research demonstrated the efficiency of CNN architecture estimation for detecting and diagnosing pneumonia using Bayesian optimization. The proposed network proved to have promising results, despite not using common preprocessing techniques such as histogram equalization and lung segmentation. This fact shows that the proposed method provides efficient and high-performance neural networks since image preprocessing is unnecessary.Recent developments of low-cost, compact acoustic sensors, advanced signal processing tools and powerful computational resources allow researchers design new scoring systems for acoustic detection of arterial stenoses. In this study, numerical simulations of blood flow inside stenosed arteries are performed to understand the effect of stenosis severity and eccentricity on the turbulence induced wall pressure fluctuations and the generated sound.
Axisymmetric and eccentric elliptic stenoses of five different severities are generated inside a 6.4 mm diameter femoral artery model. Large eddy simulations of pulsatile, non-Newtonian blood flow are performed using the open source software OpenFOAM.
Post-stenotic turbulence activity is found to be almost zero for 50 and 60% severities. For severities of 75% and more, turbulent kinetic energy rises significantly with increasing severity. The location of the highest turbulence activity on the vessel wall from the stenosis exit decreases with increasing severity.non-invasive diagnosis. Computational fluid dynamics studies that simulate large number of cases with different stenosis severities and morphologies will play a critical role in developing the necessary sound databases, which can be used to train new diagnostic devices.
Sound patterns generated from simulation results are similar to the typical sounds obtained by Doppler ultrasonography, and present distinct characters. Together with a sensor technology that can measure these sounds from within the stenosed artery, they can be processed and used for the purpose of non-invasive diagnosis. Computational fluid dynamics studies that simulate large number of cases with different stenosis severities and morphologies will play a critical role in developing the necessary sound databases, which can be used to train new diagnostic devices.To assess adherence to anti-hypertensive medication by pregnant women and to identify the factors associated with adherence or lack thereof.
Observational study in 100 pregnant women with either chronic hypertension or gestational hypertension who were being treated with at least one anti-hypertensive medication and attending antenatal clinics at one of two maternity hospitals. https://www.selleckchem.com/products/VX-680(MK-0457).html In-depth interviews were conducted with a subset of 27 women from the same group. Quotes from interview transcripts were used to illustrate the quantitative results.
BP control, self-reported adherence, complexity of medication regimen.
Participants (mean age 33 [±4.9] years; mean gestation 29 (±7) weeks) had a median blood pressure (BP) of 130/80?mmHg (IQR 16/15). Sixty-five women had chronic hypertension, of whom 13 were diagnosed during pregnancy, before 20?weeks gestation. Thirty-five women had gestational hypertension. Ninety-two per cent of participants had sub-optimal adherence. There were no significant differences in adherence scores between participants with chronic hypertension and their counterparts.